Waterjet quoting software, and the edge quality nobody specifies.

Waterjet is the one process where the same geometry, in the same material, at the same thickness, can honestly be quoted at wildly different prices, and both numbers can be right. What separates them is how good the edge has to be. That decision is usually made by the operator, after the quote went out, based on nothing anybody wrote down.

I owned and ran a machine shop for the better part of a decade before I built software for one. Every trade has one variable that quietly decides margin. On a waterjet, this is it.

Edge quality is the price lever, and customers rarely state it

Cut a part fast and the edge comes off rough, with visible striations toward the bottom and more taper. Slow the same cut down and the edge cleans up. The difference between the roughest usable finish and a finish you could hand to a customer without apologizing is not a small percentage. It is a multiple.

Here's the problem. Prints often say nothing about edge finish, so the estimator picks one in their head, and the operator picks a possibly different one at the machine. When those two picks disagree, either you gave away hours you never quoted or you shipped an edge the customer sends back. Neither shows up as a quoting problem in the numbers. It shows up as a job that ran long, or a return, and gets blamed on the floor.

The fix is not clever software. It is making the choice explicit on the quote, so the number that went to the customer and the setting used at the machine are the same decision, written down once. That is a small change and it is worth real money.

What else actually belongs in a waterjet quote

• Abrasive, as its own line. Garnet consumption per hour is often the largest running cost on the machine and the easiest one to bury inside a machine rate that hasn't been revisited. When abrasive prices move, every quote built on the old rate is quietly wrong.
• Piercing, especially on brittle and layered material. Glass, stone, composites and laminates need a gentler pierce, and gentle means slow. A part with many interior features is a different job from its perimeter alone.
• Material and thickness, because they set everything. The same machine cuts gasket material and four inch titanium. A single per-inch rate across that range is a cross-subsidy nobody chose.
• Taper, and what you're doing about it. Accepting it, slowing down for it, or using a taper-compensating head are three different costs, and they belong to the part's tolerance rather than to the material.
• Consumables and pump time as one honest rate. Orifices, mixing tubes, high pressure seals, pump maintenance, water, power, spent abrasive disposal, and the machine payment. Mixing tubes in particular wear continuously and are rarely costed per job.
• Sheet utilization, not part weight. You bought a sheet or a plate. What nests well with the rest of the order costs less than the same part standing alone.
• Part retrieval and handling. Small parts drop into the tank and somebody fishes them out. Tabbing avoids that and costs cut time. Either way it's real labor and it scales with part count.
• Secondary work. Deburr, edge finishing where the customer needs better than the cut gives, and any outside process with its lead time attached.

What I will not do: guess your cut speeds

There are platforms that read a geometry and tell you how long it will take on your machine. If that's what you're shopping for, buy one, and I mean that plainly. Speed depends on your pump pressure, your orifice and mixing tube condition, your abrasive grade and flow, your material supplier and the edge quality you actually intend to deliver. A number from somebody else's assumptions is a number your people won't stand behind.

What I'll do instead is stop your estimator re-deriving what you already know. In the build running in production today, when a part number comes up the tool reads what that part actually took on the jobs that have already run it, operation by operation, out of the ERP, matching on your part number even when the customer calls it something different. That's recall, not prediction. It can't be wrong the way an estimate can be wrong.

What a right-sized quoting tool does

Requests land in a shared mailbox and that arrival is the trigger. The system reads what it can out of the request and the attached drawings: quantities, part numbers, material and thickness, tolerance and finish callouts. It does the reading, extracting and typing so the estimator doesn't.

From there it's their call. They set the routing and the costing. Before the quote leaves, the system checks every line carries the costs it's supposed to, so abrasive and handling don't quietly go out at zero, and so the edge quality the price was built on is recorded rather than assumed.

Your machine rate, your abrasive cost, your material families, your quality levels and what each one does to speed, your minimum charges: that's the part built around how you do it. Most of a quoting tool is the same wherever it goes and is already built. The part that isn't is your pricing, and the build is sized to that.

What it doesn't do

• It's faster than the system it replaced. That's the honest claim. No percentage, because I didn't measure one.
• It doesn't nest. If you run nesting software, nesting stays there and the quote reads your assumptions.
• It won't work out your machine rate for you. It applies whatever rate you give it, consistently, which is more than most operations have today. Getting the real number is a conversation. The method is the same one in working out your wire EDM cost per hour, with your consumables swapped in.
• Reading past actuals needs a system that holds them. Proven against JobBOSS. Anywhere else is a connector question I'll answer straight rather than promise here.
• It uses an LLM to read requests and drawings. A per-use or monthly cost. Small, but real, and you should know it's there.

What it costs

A paid Diagnostic first, mapping how your quoting actually runs today. Then a Build priced up front, scoped to the part that's specific to you. I host and run it on managed infrastructure that works on top of the systems you already use, with an ongoing care plan after that keeps it running: hosting, monitoring, fixes and small adjustments as your work changes. Term and price by agreement. No hourly meter and no per-seat license.

Common questions

Most of our work is job shop cutting for other people. Does that change the build?

It changes what matters. Contract cutting lives on turnaround, on minimum charges, and on quoting the same way twice for a repeat customer. That's a smaller build than a full routing and costing system, and it should be priced like one.

We cut materials nobody else will touch.

Then your pricing logic is the asset and it almost certainly lives in one person's head. That is exactly the part worth writing down, and the reason for doing it is not the software. It's that right now the operation depends on that person being available.

We run waterjet and laser both.

Common, and they price differently enough that one set of logic can't serve both honestly. The laser note is here.

Related. The general case is quoting software specced to how you actually quote. If cutting is one step of a bigger job, see sheet metal and fabrication quoting. A quote is only as good as what the last one really took, which is why those numbers often aren't trusted.